21 Nov 2001 9:10 AR AR151-02.tex AR151-02.SGM ARv2(2001/05/10) P1: GJC Annu. Rev. Fluid Mech. 2002. 34:19–35 Copyright c 2002 by Annual Reviews. All rights reserved G.K. BATCHELOR AND THE HOMOGENIZATION OF TURBULENCE H.K. Moffatt Department of Applied Mathematics and Theoretical Physics, University of Cambridge, Cambridge CB3 9EW, United Kingdom; e-mail: [email protected] Key Words homogeneous turbulence, passive scalar problem, Kolmogorov theory, turbulent diffusion ■ Abstract This essay is based on the G.K. Batchelor Memorial Lecture that I delivered in May 2000 at the Institute for Theoretical Physics (ITP), Santa Barbara, where two parallel programs on Turbulence and Astrophysical Turbulence were in progress. It focuses on George Batchelor’s major contributions to the theory of turbu- lence, particularly during the postwar years when the emphasis was on the statistical theory of homogeneous turbulence. In all, his contributions span the period 1946–1992 and are for the most part concerned with the Kolmogorov theory of the small scales of motion, the decay of homogeneous turbulence, turbulent diffusion of a passive scalar field, magnetohydrodynamic turbulence, rapid distortion theory, two-dimensional tur- bulence, and buoyancy-driven turbulence. 1. INTRODUCTION George Batchelor (1920–2000) (see Figure 1) was undoubtedly one of the great by Cambridge University on 09/15/13. For personal use only. figures of fluid dynamics of the twentieth century. His contributions to two major areas of the subject, turbulence and low-Reynolds-number microhydrodynamics, were of seminal quality and have had a lasting impact. At the same time, he Annu. Rev. Fluid Mech. 2002.34:19-35. Downloaded from www.annualreviews.org exerted great influence in his multiple roles as founding Editor of the Journal of Fluid Mechanics, co-Founder and first Chairman of Euromech, and Head of the Department of Applied Mathematics and Theoretical Physics (DAMTP) in Cambridge from its foundation in 1959 until his retirement in 1983. I focus in this article exclusively on his contributions to the theory of turbulence, in which he was intensively involved over the period 1945 to 1960. His research monograph, The Theory of Homogeneous Turbulence, published in 1953, appeared at a time when he was still optimistic that a complete solution to the problem of turbulence might be found. During the 1950s, he attracted an outstanding group of research students, many from his native Australia, to work with him in Cambridge on turbulence. By 1960, it had become apparent to him that insurmountable math- ematical difficulties in dealing adequately with the closure problem lay ahead. As 0066-4189/02/0115-0019$14.00 19 21 Nov 2001 9:10 AR AR151-02.tex AR151-02.SGM ARv2(2001/05/10) P1: GJC 20 MOFFATT by Cambridge University on 09/15/13. For personal use only. Annu. Rev. Fluid Mech. 2002.34:19-35. Downloaded from www.annualreviews.org Figure 1 Portrait of George Batchelor by Rupert Shephard (1984). 21 Nov 2001 9:10 AR AR151-02.tex AR151-02.SGM ARv2(2001/05/10) P1: GJC G.K. BATCHELOR 21 he was to say later (Batchelor 1992), “by 1960 ... I was running short of ideas; the difficulty of making any firm deductions about turbulence was beginning to be frustrating, and I could not see any real break-through in the current publications.” Over the next few years, Batchelor focused increasingly on the writing of his fa- mous textbook, An Introduction to Fluid Dynamics (Batchelor 1967), and in the process was drawn toward low-Reynolds-number fluid mechanics and suspension mechanics, the subject that was to give him a new lease on research life in the decades that followed. After 1960, he wrote few papers on turbulence, but among these few are some gems (Batchelor 1969, 1980; Batchelor et al. 1992) that show the hand of a great master of the subject. I got to know George Batchelor in 1958, when he took me on as a new research student. Batchelor had just completed his work (Batchelor 1959, Batchelor et al. 1959) on the passive scalar problem, i.e., the problem of determining the statistical properties of the distribution of a scalar field that is convected and diffused within a field of turbulence of known statistical properties. There was at that time intense interest in the rapidly developing field of magnetohydrodynamics, partly fueled by the publication in 1957 of Cowling’s Magnetohydrodynamics. Batchelor had written a famously controversial paper, On the Spontaneous Magnetic Field in a Conducting Liquid in Turbulent Motion (Batchelor 1950a; see also Batchelor 1952b), and it was natural that I should be drawn to what is now described as the passive vector problem, i.e., determination of the statistical evolution of a weak magnetic field, again under the dual influence of convection and diffusion by a known field of turbulence. Batchelor gave me enormous encouragement and support, for which I shall always be grateful, during my early years of research in this area. My view of Batchelor’s contributions to turbulence is obviously colored by my personal interaction with him, and the following selection of what I regard as his outstanding contributions to the subject has a personal flavor. But I am influenced also by aspects of his work dating from the period 1945–1960 that still generate by Cambridge University on 09/15/13. For personal use only. hot debate in the turbulence community today. Among these, for example, is the problem of intermittency, which was first identified by Batchelor & Townsend (1949) and which perhaps contributed to that sense of frustration that afflicted Batchelor (and many others) from 1960 onward. Annu. Rev. Fluid Mech. 2002.34:19-35. Downloaded from www.annualreviews.org 2. MARSEILLE (1961): A Watershed for Turbulence These frustrations came to the surface at the now legendary meeting held in Marseille (1961) to mark the opening of the former Institut de M´ecanique Statis- tique de la Turbulence (Favre 1962). This meeting, for which Batchelor was a key organizer, turned out to be a most remarkable event. Kolmogorov was there, together with Obukhov, Yaglom, and Millionshchikov (who had first proposed the zero-fourth-cumulants closure scheme, in which so much work and hope had been invested during the 1950s); von Karman and G.I. Taylor were both there—the great 21 Nov 2001 9:10 AR AR151-02.tex AR151-02.SGM ARv2(2001/05/10) P1: GJC 22 MOFFATT father figures of prewar research in turbulence— and the place was humming with all the current stars of the subject—Stan Corrsin, John Lumley, Philip Saffman, Les Kovasznay, Bob Kraichnan, Ian Proudman, and George Batchelor himself, among many others. One of the highlights of the Marseille meeting was when Bob Stewart presented results of the measurement of ocean spectra in the tidal channel between Vancouver Island and mainland Canada (subsequently published by Grant et al. 1962). These 5/3 were the first convincing measurements to show several decades of a k spectrum and to provide convincing support for Kolmogorov’s (1941a,b) theory, which had been published 20 years earlier. But then, Kolmogorov gave his lecture, which I recall was in the sort of French that was as incomprehensible to the French themselves as to the other participants. However, the gist was clear: He said that quite soon after the publication of his 1941 papers Landau had pointed out to him a defect in the theory, namely, that wherever the local value of is larger than the mean, there the energy cascade will proceed more vigorously, and an increasingly intermittent distribution of (x,t) is therefore to be expected. Arguing for a log- normal probability distribution for , a suggestion that he attributed to Obukhov, Kolmogorov showed that the exponent ( 5/3) should be changed slightly and that higher-order statistical quantities would be more strongly affected by this intermittency. This must in fact have been no real surprise to Batchelor because, as indicated above, it was he and Townsend who had remarked on the phenomenon of intermit- tency of the distribution of vorticity in their 1949 paper, The Nature of Turbulent Motion at Large Wave-numbers. They had noticed the puzzling increase of flatness factor (or “kurtosis”) of velocity derivatives with increasing Reynolds number, a behavior that is inconsistent with the original Kolmogorov theory. They interpreted this in terms of a tendency to form “isolated regions of concentrated vorticity,” and it is interesting to note that much of the research on turbulence from the past two decades has been devoted to identifying such concentrated vorticity regions, by Cambridge University on 09/15/13. For personal use only. both in experiments and in numerical simulations. Townsend thought in terms of a random distribution of vortex tubes and sheets (Townsend 1951b) in his theory for the dissipative structures of turbulence; a theory described in Batchelor’s (1953) monograph. Annu. Rev. Fluid Mech. 2002.34:19-35. Downloaded from www.annualreviews.org I still see the 1961 Marseille meeting as a watershed for research in turbu- lence. The very foundations of the subject were shaken by Kolmogorov’s presen- tation; and the new approaches, particularly Kraichnan’s (1959) Direct Interaction Approximation, were of such mathematical complexity that it was really diffi- cult to retain that essential link between mathematical description and physical understanding, which is so essential for real progress. Given that Batchelor was already frustrated by the mathematical intractability of turbulence, it was perhaps the explicit revelation that all was not well with Kolmogorov’s theory that finally led him to abandon turbulence in favor of other fields. He had invested huge effort in the elucidation and promotion of Kolmogorov’s theory (see below) and regarded it as perhaps the one area of the 21 Nov 2001 9:10 AR AR151-02.tex AR151-02.SGM ARv2(2001/05/10) P1: GJC G.K.